System.ArgumentException occurring only in single-file published application

[francescomazzurco]

Hi,

The following line of code:

machinelearning/src/Microsoft.ML.Core/Data/Repository.cs

Line 288 in ff01708

var versionInfo = FileVersionInfo.GetVersionInfo(typeof(RepositoryWriter).Assembly.Location);

Seems to create some problems when the application is published as a single file. I honestly don't have a full understanding on how some paths and locations behave in this scenario.

For completeness, here is the exception with its stack trace:

Training failed with the exception: System.ArgumentException: The path is empty. (Parameter 'path')`
at System.IO.Path.GetFullPath(String path)
   at System.Diagnostics.FileVersionInfo.GetVersionInfo(String fileName)
   at Microsoft.ML.RepositoryWriter.CreateNew(Stream stream, IExceptionContext ectx, Boolean useFileSystem)
   at Microsoft.ML.ModelOperationsCatalog.Save(ITransformer model, DataViewSchema inputSchema, Stream stream)
   at Microsoft.ML.AutoML.ModelContainer..ctor(MLContext mlContext, FileInfo fileInfo, ITransformer model, DataViewSchema modelInputSchema)
   at Microsoft.ML.AutoML.RunnerUtil.TrainAndScorePipeline[TMetrics](MLContext context, SuggestedPipeline pipeline, IDataView trainData, IDataView validData, String groupId, String labelColumn, IMetricsAgent`1 metricsAgent, ITransformer preprocessorTransform, FileInfo modelFileInfo, DataViewSchema modelInputSchema, IChannel logger)
   at Microsoft.ML.AutoML.Experiment`2.Execute()
   at Microsoft.ML.AutoML.ExperimentBase`2.Execute(ColumnInformation columnInfo, DatasetColumnInfo[] columns, IEstimator`1 preFeaturizer, IProgress`1 progressHandler, IRunner`1 runner)
   at Microsoft.ML.AutoML.ExperimentBase`2.ExecuteCrossValSummary(IDataView[] trainDatasets, ColumnInformation columnInfo, IDataView[] validationDatasets, IEstimator`1 preFeaturizer, IProgress`1 progressHandler)
   at Microsoft.ML.AutoML.ExperimentBase`2.Execute(IDataView trainData, ColumnInformation columnInformation, IEstimator`1 preFeaturizer, IProgress`1 progressHandler)
   at Microsoft.ML.AutoML.ExperimentBase`2.Execute(IDataView trainData, String labelColumnName, String samplingKeyColumn, IEstimator`1 preFeaturizer, IProgress`1 progressHandler)

Thanks

[michaelgsharp]

How are you publishing everything as a single file? Are you using something to embed all the resources into a single .dll?

[francescomazzurco]

Yes, I am using the option <IncludeNativeLibrariesForSelfExtract>True</IncludeNativeLibrariesForSelfExtract> in my PublishProfile in order to get a single file.

[pplusawork]

It looks like this is related to #4505 and PR #4512 which looks like it was blocked pending review due to code coverage?

In my case, I was using a self-contained Linux-x64 publish profile, and checking the "Produce single file" box, i.e. <PublishSingleFile>True</PublishSingleFile>. Setting that to false seems to be a reliable work around, though not ideal.